You should enable Bidirectional Forwarding Detection (BFD) to try to reduce the time to detect a failed link to less than one second. BFD is a detection protocol that is designed to detect forwarding path failures at a consistent rate, thereby providing network administrators with predictable reconvergence times.

Additionally, BFD is designed to work regardless of media type, encapsulation, or routing protocol, providing network administrators with a uniform forwarding failure detection method across a network. BFD supports Open Shortest Path First (OSPF), Enhanced Interior Gateway Routing Protocol (EIGRP), Border Gateway Protocol (BGP), and Intermediate SystemtoIntermediate System (IS-IS).

The detection of a forwarding path failure causes BFD to notify the routing protocol that a link has failed, which causes the routing protocol to recalculate the routing table. BFD works by sending control packets between two adjacent routers to create a BFD neighbor session. Once the neighbor relationship is established, the two adjacent routers send control packets to each other to maintain the neighbor relationship, similarly to how routing protocols maintain neighbor relationships.

However, BFD sends packets at a much faster rate than routing protocols do. In addition, BFD can distribute some functionality from the control plane to the data plane, thereby requiring fewer CPU resources than routing protocol timers do. Only a single BFD session is established per interface regardless of how many routing protocols are running on that interface. You should not enable Spanning Tree Protocol (STP) UplinkFast to try to reduce the time to detect a failed link to less than one second. STP UplinkFast is a Ciscoproprietary STP convergence enhancement. You can enable STP UplinkFast to reduce STP convergence time from the standard 14to30 seconds down to one second, but not to less than one second.

You should not implement OSPF fast hellos to try to reduce the time to detect a failed link to less than one second. OSPF fast hellos can send multiple hello packets per second, which results in a faster convergence time. However, the detection and notification of link failures also depend on the OSPF dead interval, which can be set to a minimum of one second.

You should not reduce OSPF hello timers to try to reduce the time to detect a failed link to less than one second. Although reducing the OSPF dead interval and increasing the number of hellos sent during the dead interval can result in hellos being sent at a subsecond rate, the OSPF dead interval can be set to a minimum of one second; therefore, reducing OSPF hello timers can reduce the time to detect a failed link to a minimum of one second.

You should not switch to EIGRP and reduce hello and hold timers to try to reduce the time to detect a failed link to less than one second. Reducing EIGRP timers can reduce the time to detect a failed link to less than two seconds.